bounded rationality and criminal investigations: has ... · pdf filebounded rationality and...
TRANSCRIPT
Bounded Rationality and Criminal Investigations
1
Bounded Rationality and Criminal Investigations:
Has Tunnel Vision Been Wrongfully Convicted?
Brent Snook and Richard M. Cullen
“Cognition is the art of focusing on the relevant and deliberately ignoring the rest.”
Gerd Gigerenzer and Peter Todd
A substantial portion of judgment and decision making research has led to the conclusion
that using heuristics – simple mental strategies that people use to deal with our uncertain world –
results in erroneous decisions. The message that “heuristics are bad” primarily stems from a
wealth of research showing that human decision-making deviates from idealistic, statistics-based
decision-making processes that strive for optimality (Kahneman, Slovic, & Tversky, 1982;
Nisbett & Ross, 1980). In particular, it has been noted that heuristics ignore apparently relevant
information, whereas the idealistic models are thought to examine everything. The negative
view of heuristics has spread to many domains (see Gilovich, Griffin, & Kahneman, 2002, for
some examples), including criminal investigations where the use of heuristics by police officers
is thought to produce reasoning errors that contribute to criminal investigative failures (e.g.,
Findley & Scott, 2006). One heuristic-like process that is cited frequently as an explanation for
criminal investigative failures is “tunnel vision.” If investigating officers, for example, stop
searching for additional suspects after locating a viable suspect, they may be accused of using
tunnel vision. Despite a complete absence of empirical research on tunnel vision in criminal
investigations, there have been calls to eradicate this mental “virus” (e.g., Cory, 2001; Findley &
Scott, 2006). Specifically, it has been recommended that police officers should avoid using
Bounded Rationality and Criminal Investigations
2
tunnel vision by employing more deliberate and careful decision-making strategies. While this
solution is intuitively appealing, its feasibility is questionable given: (1) the constrained context
of criminal investigative decision making; and (2) the processing limitations of the human mind.
In this chapter, we outline a psychological framework called bounded rationality and
illustrate how it applies to investigative decision making. Applying the bounded rationality
perspective involves taking an ecological view of cognition by outlining the actual context where
police officers work and determining whether the heuristics that officers use are efficient and
effective decision-making strategies within that context. In taking an ecological view, we hope
to gain some insight about when and why heuristics are likely to succeed and fail in the criminal
investigative environment. We use tunnel vision as a primary example of how heuristics in
policing have been vilified (see Lerner, 2005, for a more detailed discussion of how police
heuristic-led judgments, are criticized in the criminal justice system). Tunnel vision appears to
consist of a set of heuristics, which are arguably adaptive mechanisms that have evolved in the
mind to allow people to make smart decisions. As with all judgments and decisions, decisions
made at various points in the investigative process are constrained by time, knowledge, and
mental capacity. We believe that it is unrealistic to expect police officers to investigate all
possible suspects, collect evidence on all of those suspects, explore all possible avenues
concerning the circumstances surrounding a crime, search for disconfirming and confirming
evidence of guilt for every suspect, and integrate all of this information to make an “optimal”
decision.
Has Tunnel Vision been Wrongfully Convicted as a Flawed Mental Tool?
Bounded Rationality and Criminal Investigations
3
Cases of wrongful conviction are being uncovered at an increasing rate and have
rightfully received much public scrutiny (Huff, 2004; Huff, Rattner, & Sagarin, 1986; Rosen,
1992; Scullion, 2004). Such cases have devastating effects on wrongfully convicted individuals
(see Campbell & Denov, 2004; Grounds, 2004) and allow guilty offenders to go free, thereby
bringing disrepute and public mistrust to the administration of justice. In recognition of the need
to prevent wrongful convictions, the Canadian Federal-Provincial-Territorial Heads of
Prosecutions Committee (hereafter referred to as the FPT Committee) set up a Working Group
on the Prevention of Miscarriages of Justice in 2002 to identify the factors that contribute to
these justice system errors. The mandate of the FPT Committee was to, amongst other goals,
ascertain why wrongful convictions were occurring, how criminal investigations were failing,
how police resources could be used more efficiently, and how to facilitate the timely resolution
of cases. The FPT Committee concluded that criminal investigative failures were sometimes a
function of unethical conduct by investigators who assigned blame to the wrong individuals. In
addition, the FPT Committee concluded that investigators sometimes failed to use best practices
(e.g., having knowledge about recent research on eyewitness identification and testimony, line-
up methods, interviewing and interrogation strategies, jail-house informants, and DNA
technology), and that investigators suffered from tunnel vision. According to the varied
definitions that have been offered, tunnel vision in the criminal investigative context involves:
(1) identifying a primary suspect; (2) searching for information about that suspect; and (3)
ignoring information that might disconfirm that the primary suspect is the culprit, including
information about other plausible suspects.
The FPT Committee provided a series of policy recommendations aimed to eliminate, or
at least reduce, future miscarriages of justice. They recommended that police agencies should
Bounded Rationality and Criminal Investigations
4
implement training, screening, and disciplinary policies to deal with unethical conduct; police
officers should be educated on best practices; and that police officers should avoid tunnel vision.
Although we wholeheartedly agree with the first two recommendations, we take issue here with
the last one.
Those who argue that tunnel vision is a cause of wrongful convictions seem to believe
that bad outcomes (the conviction of an innocent suspect) only result from either bad decision-
making processes or bad investigators. But good processes and good investigators can also be
associated with bad outcomes. Heuristics are normally effective and efficient strategies for
handling complex information and drawing conclusions from that information, but in some
instances can lead to error. The heuristics that make up tunnel vision are no exception. For
example, even the most decorated police officer can be led astray by “misleading information”
such as a fabricated eyewitness account (although this would not be known to that officer until
after the fact). And whereas bad (e.g., malicious, indifferent, or “nobly corrupt”) investigators
may indeed be the cause of some investigations going awry, tunnel vision is an altogether
different process.
The recommendation to “avoid,” “correct,” or “prevent” tunnel vision is therefore
premature. Not enough is known about tunnel vision to make such recommendations. More
specifically, such a recommendation is as likely to be ineffective as it is to be effective because:
(1) tunnel vision is an ambiguous concept; (2) there has been no systematic study of the
proportion of successful cases where police officers used tunnel vision; and (3) there has been no
proper evaluation of the contribution of tunnel vision to wrongful convictions. Given that
current complaints about tunnel vision are based on retrospective analysis of investigative fiascos
(Findley & Scott, 2006) and the lack of controlled experimental research on the topic, it is not
Bounded Rationality and Criminal Investigations
5
surprising that there is no compelling empirical evidence to support the message that tunnel
vision is a bad decision-making strategy. Indeed, the recommendations to correct tunnel vision
appear to be based on nothing but “bad common-sense reasoning” (see Gendreau, Goggin,
Cullen, & Paparozzi, 2002, for how bad common sense based policy recommendations, as
opposed to those based on empirical evidence, can lead to the implementation of ineffective
policies).
The idea that police officers should be wary of tunnel vision mirrors an ongoing debate in
psychology about human rationality. Policy-makers and researchers who have prematurely
focused upon tunnel vision as a flawed mental strategy might be able to increase the likelihood
of reducing the occurrence of investigative failures by considering the issues that are at the heart
of this debate, particularly the arguments that have been put forth since bounded rationality
theory originated in the 1950s. Consequently, the primary goal of this chapter is to expose
readers to the relatively recent developments in the wider rationality debate and illustrate how
this debate is applicable to the understanding of heuristic-led judgments in criminal
investigations.
We begin with an overview of the rationality debate. Put simply, researchers on one side
focus disproportionately on the instances where heuristics produce errors. These researchers
argue that using heuristics is irrational because heuristics are suboptimal to complex decision-
making models that supposedly define the best possible way to make decisions. The other side
argues that heuristics lead to good decisions. According to this second view, tunnel vision might
be helpful to police officers on a psychological level, for example, by allowing them to focus
their thoughts in a complex investigative environment. We then describe the criminal
investigative environment and argue that it is unrealistic to expect officers to use what are
Bounded Rationality and Criminal Investigations
6
commonly referred to as fully rational decision-making models. This will be followed by an
attempt to operationalize tunnel vision using existing heuristics that have been outlined and
tested in the psychological literature.
The Rationality Debate in Psychology
The rationality debate is primarily about whether people make good decisions. Arguably,
the most contentious issue in this debate is about how to best measure good decisions. Over the
years, psychologists have varied the decision-making benchmark between how people perform in
the real world to achieve their goals and objectives (referred to as “rationality1” by Mantkelow,
1999) and whether people live up to normative standards (referred to as “rationality2” by
Mantkelow, 1999). In order to be judged rational2, a person would have to search indefinitely
for endless amounts of information, have knowledge of every relevant aspect, weigh all the
available information according to importance, and finally perform intractable mathematical and
statistical calculations. (Such a person has been called homo economicus, or economic man.) If
rationality is thought to be synonymous with optimality – which has often been the case – then
unbounded models of this sort become the definition of rational thinking. People appear doomed
to be irrational if such an unattainable standard is maintained.
Bounded rationality researchers have challenged the view of the human mind as a
statistical software package (Gigerenzer, 2000; Kahneman & Tversky, 1973; Simon, 1955, 1956)
by arguing that the natural constraints on decision making must be taken into account when
assessing human rationality. These researchers proposed that heuristics are more
psychologically plausible models of human decision making than those that involve a high
degree of information integration and abstract mathematical assignments and calculations. Most
Bounded Rationality and Criminal Investigations
7
recently, Gerd Gigerenzer and his colleagues (e.g., Gigerenzer, Todd, & The ABC Research
Group, 1999; Gigerenzer, 2001; Todd & Gigerenzer, 2000) have shown that heuristics are
rational because they work well in natural environments.
We therefore adopt the rationality1 definition, and measure rationality by how well
people make decisions in the real world, where decision making is limited by available time,
information or knowledge, and mental resources.
Historical Developments
While it is generally acknowledged that people use heuristics to make all sorts of
decisions, the debate continues over whether their use leads to good or bad decisions. There
were three especially influential contributions to this debate in the second half of the last century
(see Gilovich & Griffin, 2002; Goldstein & Hogarth, 1997). First, Paul Meehl (1954)
empirically compared the judgments of expert clinicians (which were presumably made with
heuristics) with judgments reached by actuarial models and made two important discoveries.
First, the actuarial methods almost always outperformed the experts. Second, the experts thought
they performed better than they actually did. These two findings inspired further research on
ways that the mind produces convincing but erroneous judgments. This further research
generally supported Meehl’s conclusions (e.g., Goldberg, 1965; Sawyer, 1966) that humans
cannot reason as well as they should be able to.
Nobel Prize winning economist and cognitive scientist, Herbert Simon, refuted this
conclusion because it suggested that humans are irrational. Simon thought that it might be the
definition of rationality that was the problem and not human performance on judgment and
decision-making tasks. Simon is now credited with developing the argument that the level of
Bounded Rationality and Criminal Investigations
8
rationality2 suggested by rational choice models was an impractical standard for assessing human
judgment and rationality. Proponents of rational choice models assume that people make
judgments by evaluating the probability of each outcome, determining the utility to be gained
from each outcome, combining these two evaluations, and then choosing the option that offers
the optimal combination of probability and utility (Gilovich & Griffin, 2002). Simon (1955,
1956) was opposed to this classical criterion of full rationality, and proposed the principle of
bounded rationality as a more realistic standard. According to Simon’s view, judgment and
decision making is constrained by the processing limitations of the mind so humans have
evolved to use “simple” strategies that can handle complex information (see also Kahneman,
1973; Miller, 1956). Simon believed that humans are in fact rational considering that they make
decisions under limited mental capacity and within the complexities of our uncertain world. He
compared heuristics to a pair of scissors, where one blade represents the environment and the
second represents the mind. Simon asserted that one must study how the two fit together, like
the blades of the scissors, in order to appreciate how heuristics work.
The third important contribution to the rationality debate was the inception of Bayesian
statistical analyses, a form of probability calculus, into the field of psychology by Ward Edwards
(Edwards, Lindman, & Savage, 1963). The Bayesian models set a normative standard that
heuristic-led judgments could be compared to (e.g., Edwards, 1968). Normative standards are
those that establish how things should or ought to be, which things are good or bad, or which
actions are right or wrong. Through such comparisons, it appeared that heuristic-led judgments
were less than optimal because people never performed the way they ought to according to the
idealistic benchmarks. An interest in the causes of this shortcoming subsequently emerged, as
Bounded Rationality and Criminal Investigations
9
well as interest in determining ways to correct the apparent “flaws” of human cognition
(Gilovich & Griffin, 2002).
Heuristics and Biases
Following these developments, Daniel Kahneman and Amos Tversky published a series
of papers about how heuristics can sometimes lead to errors and biases (e.g., Kahneman &
Tversky, 1972, 1973; Tversky, 1972; Tversky & Kahneman, 1971, 1973; 1974; 2002). A
collection of their works, along with articles of other like-minded researchers, appeared in a now
classic book titled Judgment Under Uncertainty: Heuristics and Biases (Kahneman, Slovic, &
Tversky, 1982). Its main message is that people often use heuristics rather than fully rational
models to make judgments under uncertainty. The contributors proposed that heuristics can
yield both good and bad decisions, challenged whether complex normative models of human
judgment accurately described underlying mental processes, and attempted to explain the range
of observed human errors as the systematic result of cognition without implying that humans are
irrational (Gilovich & Griffin, 2002). This program of research became known as the “heuristics
and biases” program.
They discovered that everyday judgments do not adhere to the laws of probability or to
statistical principles and argued that the underlying processes in decision making were altogether
different than those implied by rational choice models. They subsequently proposed that people
employ a limited number of simple cognitive rules, or heuristics, that evaluate the likelihood of
options using basic computations that the mind can perform. They proposed three judgmental
heuristics – the representativeness heuristic, the availability heuristic, and the anchoring and
adjustment heuristic – that are commonly used to estimate probabilities, frequencies, and values,
Bounded Rationality and Criminal Investigations
10
are cognitively cheap, and are usually effective (see Chapter 2 of this book for a discussion of
how these heuristics have been observed in criminal investigative failures). Heuristics were
defined as any automatic or deliberate strategy that uses a natural assessment in order to estimate
or predict something. The representativeness heuristic, for instance, involves the classification
of things based on how similar they are to a typical case. It is supposedly used when trying to
determine the probability that object A belongs to class B. The subjective probability judgment
rests on how representative object A is of class B. To use a criminal investigative example,
when inferring whether a particular person is likely to be guilty, police officers might mentally
compare the suspect to their perception of a prototypical offender. If the suspect does not show
remorse, for example, a police officer might be inclined to believe the suspect is guilty (see
Weisman, 2004, for a discussion of how showing remorse is interpreted by officials in the
criminal justice system).
Kahneman and Tversky argued that biases occur because heuristics denote a tendency to
make a choice that is inaccurate. For instance, the representativeness heuristic could yield an
incorrect judgment if a suspect did not show remorse but was actually innocent, because a lack
of remorse may not always indicate guilt (an innocent suspect might not show remorse). It was
the tendency for different people to make remarkably similar errors on similar tasks, relative to
the normative models, that led to the conceptualization of the three aforementioned judgmental
heuristics. The predictability of the biases invoked research into the cognitive mechanisms that
caused them – heuristics. However, the biases continued to receive most of the scholarly
attention in the immediate years to follow. Although it was apparently not Kahneman and
Tversky’s intention (see Gilovich & Griffin, 2002), the disproportionate focus on the instances
where heuristics lead to error, rather than the instances where they lead to good decisions,
Bounded Rationality and Criminal Investigations
11
combined with the continued scholarly acceptance that normative models were the most superior
method of making decisions, appears to have produced the belief that heuristics are bad, a belief
that still exists today. A negative image of human cognition was thus cast and the “cognitive
miser” image was born (Fiske & Taylor, 1991). According to this image, humans are thought to
deliberately sabotage their own accuracy by using heuristics because they are too lazy (or cheap)
to carry out extensive computational processes. Research illustrating the fallibility of heuristics
has now gained a strong foothold in many areas, including economics, medicine, politics, sports,
and justice (see Myers, 2002, and Piattelli-Palmarini, 1994, for a list of the many documented
heuristic-led biases).
The ABCs of Bounded Rationality
In recent years, Gigerenzer and his colleagues (e.g., Gigerenzer et al., 1999; Gigerenzer
& Selten, 2001; Gigerenzer & Todd, 1999; Todd & Gigerenzer, 2003) at the Centre for Adaptive
Behavior and Cognition at the Max Planck Institute for Human Development (hereafter referred
to as the ABC Research Group) have been challenging the unbalanced view that heuristics are
bad. Gigerenzer and Todd (1999) claim there is an unquestioned assumption in much of
psychology “that the more laborious, computationally expensive, and nonheuristic the strategy,
the better the judgments to which it gives rise” (p. 20, italics added). Those who compare human
reasoning to the unrealistic benchmarks set by rationality2 promote this “more-is-better
ideology.” The ABC Research Group do not believe that more is always better; in fact, they
have argued that less is more in certain situations (Goldstein & Gigerenzer, 1999, 2002; Todd &
Gigerenzer, 2003). In a compilation of their experimental findings and theoretical essays, titled
Simple Heuristics that Make us Smart, Gigerenzer et al. (1999) maintain that the image of
Bounded Rationality and Criminal Investigations
12
humans as irrational, resulting from years of comparing human rationality to normative models,
can be mended by considering the real and inherently uncertain environments in which people
make decisions. Essentially, the ABC Research Group maintains that heuristic reasoning
strategies have evolved over time not as suboptimal decision-making strategies, but as effective
strategies that we can use to make everyday judgments and decisions in a complex world.
Two of the ABC Research Group’s core concepts – bounded rationality and ecological
rationality – capture their central ideas. Bounded rationality originated with Simon’s (1955,
1956) notion of satisficing, which involves the mental or physical search through a series of
alternatives until one is found that meets a certain pre-defined level – called the aspiration level.
If you were searching for a house, for instance, you may decide you want a clean house in a
suburban area that is below $300,000. It is possible that you would satisfice when choosing a
house to buy because it is near impossible to look at all available houses everywhere and then
select the best option. This means you would probably buy the first house that met your
aspiration level.
Fundamental to the ABC Research Group’s bounded rationality theory, and the most
intriguing contribution to the ongoing debate about the validity of heuristics, is a metaphor which
views the mind as an adaptive toolbox. Like a carpenter’s toolbox, the mind is equipped with a
repertoire of simple mental tools that are specially suited for certain judgments and decisions.
These mental tools are fast and frugal heuristics that have evolved to allow people to make smart
decisions. The heuristics are fast because they do not involve much calculation or integration of
information, and frugal because they ignore some of the available information, thus sparing
mental resources.
Bounded Rationality and Criminal Investigations
13
The simplest tool in the adaptive toolbox is the recognition heuristic, which leads people
to choose something they recognize over something they do not recognize (Goldstein &
Gigerenzer, 1999). As an example of how the recognition heuristic might be used to make a
decision, consider this question: Which of these two National Hockey League players has
achieved the highest total career points – Mark Messier or Eric Cairns? If you only recognize
one player and not the other, you will use the recognition heuristic. Did you choose Mark
Messier? Was it because you recognized Messier and not Cairns? If so, you made a correct
inference by using the recognition heuristic (see Snook & Cullen, 2006). Given a set of options,
the heuristics in the adaptive toolbox specify how people search through the attributes that are
associated with the options, stop that search, and then make a choice. From this toolbox
perspective, human decision making is adaptive because the mind is equipped with heuristics
that meet the demands of a variety of decision tasks.
Ecological rationality is concerned with the structure and representation of information
in the environment and how well heuristics match that structure. To the extent that such a match
exists, heuristics allow people to make an accurate decision quickly (i.e., the heuristic is
ecologically rational). By focusing on the match between the environment and the mind, the
ABC Research Group have placed human reasoning into an evolutionary framework that is
omitted from most decision-making theories. They do not define errors by how far the outcome
and process deviate from rules specified by rational choice models. By contrast, they consider
the ecological rationality of a strategy to assess whether it is effective in a particular situation.
To continue with our hockey player example, the recognition heuristic is ecologically rational for
this particular decision because good hockey players are more recognizable than bad hockey
players. In addition to receiving media attention for being a good player, Messier also received
Bounded Rationality and Criminal Investigations
14
wide media exposure through his endorsement of Lays Potato Chips. Of course, Lays would not
have hired an unrecognizable player in the first place. Essentially, people are able to capitalize
on the fact that media exposure is a reflection of hockey greatness because the best players
receive relatively more media exposure, and thus have a greater likelihood of being recognized.
Bounded rationality is based on the premise that our minds construct simplified models
of the complex world in order to deal with uncertainty. The performance of these heuristics has
been compared to complex methods in a series of studies. In perhaps the most comprehensive
study, Czerlinski, Gigerenzer, and Goldstein (1999) compared the performance of simple
heuristic models against multiple regression – a complex statistics-based model – in 20 different
decision environments (e.g., predicting average attractiveness ratings of famous men and
women). They found that the heuristic models provided an equally good fit to a range of data
sets and tend to do so with fewer cues (i.e., they are more frugal). Similar results have been
reported by Dhami and Harries (2001) in their study of how a group of general practitioners
would decide to prescribe blood pressure medication, by Snook, Taylor, and Bennell (2004) in
their study of how people predict offender home locations, by Dhami (2003) in her study of how
judges make bail decisions, and by Smith and Gilhooly (2006) in their study of practitioners’
decisions to prescribe anti-depressant medication. Taken together, these studies have shown that
the fast and frugal heuristic models provide a psychologically plausible account of how people
make all sorts of judgments and decisions.1
1 Other researchers have also suggested that research on cognitive accomplishments has been
“crowded out” by research on cognitive errors, and that statistical analyses typically focus on bias to the exclusion of accuracy. Krueger and Funder (2004), for example, argue that many “biases” can be beneficial and that when an analysis stops without asking “why” such a behavioural or cognitive tendency exists, or what general purpose it might serve, the development of integrative theory and sensible advice is stymied.
Bounded Rationality and Criminal Investigations
15
In sum, the rationality debate has a long history and is deeply entrenched in the field of
psychology. Some researchers examine rationality by comparing human reasoning to lofty
benchmarks that they believe people ought to achieve and, because people fall short of these
normative benchmarks, conclude that cognition is flawed and people are unavoidably irrational
(see Kahneman, Tversky, & Slovic, 1982; Piattelli-Palmarini, 1994). The natural response to
this observed “irrationality” has been to prescribe corrective procedures to allow people to get
closer to the benchmark. By contrast, the bounded rationality perspective is concerned with
describing how decision-making strategies allow people to function in the real world. Bounded
rationality researchers use ecological standards (accuracy, speed, frugality), rather than
normative standards, to evaluate human rationality. By considering the nature of the situation in
which a decision is made, it is possible to gain an understanding of when and why a particular
heuristic is likely to succeed or fail in that situation.
Components of the rationality debate are clearly evident in the recent criminal justice
literature that cites tunnel vision as a flawed mental process that produces criminal investigative
failures. The basis of the specific arguments that tunnel vision is a mental virus is the same as
that for the argument that heuristics lead to poor decisions – these strategies are too simple or
they ignore information. Furthermore, in the current concern about tunnel vision, as well as in
the broader debate, there has been a recognized need for corrective measures against heuristic
use. Because the bounded rationality perspective has provided some insight for the greater
rationality debate, it can also shed light on the way tunnel vision is currently viewed by criminal
justice professionals.
Bounded Rationality and Criminal Investigations
Bounded Rationality and Criminal Investigations
16
Police officers work in an environment where they are expected to be fully rational.2 This
is especially the case when investigative failures come under direct public and legal scrutiny.
When searching for suspects or through information about a set of existing suspects, police
officers are expected to investigate all possible evidence and all possible suspects, explore all
possible avenues concerning the circumstances surrounding a crime, search for disconfirming
and confirming evidence, and make an optimal decision based on the information found (Forst,
2004; Goff, 2001; Innes, 2002). These expectations are similar to those placed on human
decision making by proponents of rational choice models that assume that people have infinite
time and ability to acquire and process all the information relevant to a particular decision. As
previously mentioned, setting implausible information processing expectations can lead to the
conclusion that the decision maker, in this case the police officer, is irrational, lazy, or used a
flawed mental strategy. In other words, the expectation of optimal processing makes it seem like
police officers are not doing their jobs properly.
According to the bounded rationality perspective, however, people always use heuristics
to make decisions. To apply the bounded rationality framework to police decision making, one
must consider whether a particular heuristic meets the demands of the criminal investigative
environment.
The Bounded Investigative Environment and the Mind
The criminal investigative environment can be best characterized as a naturalistic
decision setting. Such settings typically involve time pressures, high stakes, experienced
decision-makers, inadequate (e.g., missing or uncorroborated) information, ill-defined goals,
2 Or at least parrot the overtones of full rationality (Lerner, 2005).
Bounded Rationality and Criminal Investigations
17
poorly defined procedures, stress, dynamic conditions, team coordination, interruptions,
distractions, noise, and other stressors (see Klein, 2001; Orasanu & Salas, 1993). When a crime
is reported, police officers begin a search for information to identify and locate a primary suspect
through physical (e.g., canvassing, interviewing witnesses), mental (e.g., linking related cases),
and/or archival (e.g., searching police files) sources (de Poot & van Koppen, 2004; Sanders,
1977; Innes, 2002).3 In a world without limits, an officer could conduct an infinitely large search
of all information available in the universe. In reality, however, police officers do not have the
luxury of unlimited search time. There are limitations, for example, on how many houses can be
canvassed, how much comparative analysis can be done, and how much effort can be spent
searching police records.
Criminal cases become harder to solve with time (Keppel & Weis, 1994; Mouzos &
Muller, 2001), so many investigations are a race against the clock. Time, therefore, is the first
major constraint on police decision making. Time constrains the search for information by
influencing how resources are allocated, most notably the manpower required to manage
investigative teams, interview witnesses, interact with other agencies, organize information
coming into the investigation, respond to the media, and follow lines of inquiry (Eck, 1979).
Police officers simply do not have time to search for all information that is relevant or necessary
to make an optimal decision. In the end, time-limited searches influence the quality and quantity
of information that is collected, organized, and processed.
The resources that are available during an investigation are also limited. Resource
allocation in a police agency must be prioritized to ensure that all important functions of the
agency continue to operate properly. There are limited financial resources available, and a
3 See Maguire (2003) and de Poot and van Koppen (2004) for a discussion of how different types
of crimes demand different search strategies that vary in complexity.
Bounded Rationality and Criminal Investigations
18
balance must be struck between, for example, personnel (e.g., overtime), equipment (e.g.,
radios), and new technology (e.g., forensic capabilities). Although agencies can sometimes
obtain new resources at the start of a major investigation, these cannot be sustained indefinitely.
As with time, resource limitations constrain the search for information that is used to make
investigative decisions.
Similarly, there are limitations on cognitive ability, or constraints on the mental
processing of information (e.g., Kahneman, 1973; Miller, 1956), that limit the decisions made by
investigators. At the most basic level, information processing involves encoding, storing, and
recalling information (Atkinson & Shiffrin, 1968). In order for relevant and novel information to
be encoded and stored in the first instance, one must pay attention to that information. If
attention is interrupted, by distraction for example, the encoding process can be disrupted and the
information will not become stored in memory. In addition to inattention, there are a range of
other limitations on information processing. For example, people can only hold an average of
seven pieces of information in their short term, working memory store at any given time
(Baddeley, 1992; Miller, 1956). The human mind, therefore, does not have the capacity to
consider every piece of information, weight the importance of each piece of information, and
integrate the information in a computationally expensive way. It is unrealistic, then, to expect a
police officer’s mind to act like a computer processor.
Nor should we expect police officers to have unlimited knowledge about every aspect of
criminal investigations or have access to all of the information that is required to make a perfect
decision. It is clearly impossible for police officers, or anyone else for that matter, to be fully
familiar with, aware of, or completely understand criminology, forensics, psychology, law,
biology, sociology, anthropology, linguistics, chemistry, statistics, ethics, politics, management,
Bounded Rationality and Criminal Investigations
19
and other areas that could be required to effectively investigate crimes. Knowledge is gained
through experience or study. Police officers accumulate their criminal investigative knowledge
through experience gained from working a range of different cases, through training and self-
study, and by interacting with and listening to other investigators.
The decisions detectives reach are also contingent upon both the quantity and the quality
of information about the current investigation. Police officers, for example, may have only one
eyewitness account of a crime, but must use that limited, and often ambiguous, information to
make investigative decisions (see Manning, 1977; Sanders, 1977). Even if numerous
eyewitnesses come forward, the knowledge they provide is still limited by the fallibility of
human memory (Sporer, 1996).
In addition to the four major constraints, police officers experience pressure from internal
and external sources that can influence the types of decision-making strategies that work well in
the investigative environment (Findley & Scott, 2006; Maguire, 2003). For example, Crego and
Alison (2004) conducted electronic focus groups with 28 senior investigating officers (SIOs) in
the United Kingdom for a range of different types of serious inquiries (e.g., child abductions,
stranger rapes). The officers reported the types of issues they believed were important features
of the critical incidents they had managed. Of the wide range of issues the SIOs reported having
to deal with, several of them were identified as placing considerable stress on the inquiry
process. The officers felt pressure because of the consequential nature of the decisions they
make that can affect people’s lives, the complexity of the case, the local community’s concerns,
the involvement and scrutiny of the press, the inability to influence other individuals that can
have an impact on the investigation, the managing of the investigative team, the ever-changing
nature of the investigation, and their belief that they will be blamed for anything that goes wrong.
Bounded Rationality and Criminal Investigations
20
Police are, furthermore, expected to deal with these pressures amid the regular constraints of the
political and legal systems (Lyons & Truzzi, 1991; Goff, 2001; Young, 1996).
Now that the constraints on investigative decision making have been highlighted, it is
necessary to develop a clear definition of tunnel vision in order to determine whether it is likely
to be an effective strategy in the investigative environment.
Tunnel Vision: Narrow-Mindedness or Focused Determination?
In biological terms, tunnel vision refers to a reduced field of vision – as from within a
tunnel looking out – that results from the loss of peripheral vision with retention of central vision
(Williams, 1985). A Google search of “tunnel vision” retrieves over 14 million hits, however,
and a quick scan of these shows that the term is often used metaphorically to describe how
people in a range of domains (e.g., medicine, politics, law, and business) search for information.
Tunnel vision in this sense is generally seen as a negative or undesirable process, as it refers to
the narrow-minded pursuit of evidence that supports a decision that has already been made while
ignoring evidence that may contradict that initial decision. The main argument against tunnel
vision is that some of the ignored evidence may be valid and people should therefore allow their
beliefs to change in accordance with that evidence. Like the proponents of rationality2, some
criminal justice pundits appear to believe that a different investigative process, involving the
collection and consideration of all available evidence, is superior to this “narrow-mindedness.”
But tunnel vision can also be cast in a positive light; it can be interpreted as an adaptive process
whereby people focus on important, relevant evidence while ignoring evidence that distracts
them from making a decision and acting upon it (Lewicka, 1998). This optimistic view implies
“focused determination,” a desirable human characteristic. Regardless of whether one takes a
Bounded Rationality and Criminal Investigations
21
pessimistic or an optimistic stance, the metaphorical tunnel vision appears to be referring to a
cognitive strategy, albeit an ambiguous one, that guides the search for alternatives and
information.
Tunnel vision has become a convenient catchphrase in the field of criminal justice to
refer to how the police locate suspects and build a case against them. Published articles, books,
and judicial inquiries about wrongful convictions in Canada, such as the Donald Marshall, David
Milgaard, Wilbert Coffin, Steven Truscott (Anderson & Anderson, 1998), Guy Paul Morin
(Anderson & Anderson, 1998; Epp, 1997), and Thomas Sophonow (Anderson & Anderson,
1998; Cory, 2001; Wilson, 2003) cases all cite police or prosecutorial “tunnel vision” as a
contributing factor. In addition, the Canadian Journal of Criminology and Criminal Justice
dedicated a special issue in 2004 to wrongful convictions, and the FPT Committee, discussed
earlier in this chapter, identified “tunnel vision” as one of the contributors to wrongful
convictions.
Although these different published works defined tunnel vision in slightly different ways,
most of the definitions assume that tunnel vision is a form of police misconduct. Even authors
who have attempted to define tunnel vision objectively, by referring to it as the natural way of
processing information that is gathered during an investigation (e.g., Findley & Scott, 2006;
Wilson, 2003), have still cast tunnel vision in a negative light by implying that it is a suboptimal
or flawed investigative strategy. No one appears to have considered that tunnel vision might be a
cognitive strategy that serves a useful purpose for criminal investigators. Below are some of the
definitions that have been provided, which illustrate two things: (1) a lack of agreement about
what exactly tunnel vision is; and (2) the confounding of police misconduct with systematic
cognitive processing.
Bounded Rationality and Criminal Investigations
22
Perhaps the most widely cited definition of tunnel vision is that provided by Justice Cory
in relation to the wrongful conviction of Thomas Sophonow (Cory, 2001). Cory stated that:
Tunnel vision is insidious. It can affect an officer or, indeed, anyone involved in
the administration of justice with sometimes tragic results. It results in the officer
becoming so focussed upon an individual or incident that no other person or
incident registers in the officer’s thoughts. Thus, tunnel vision can result in the
elimination of other suspects who should be investigated. Equally, events which
could lead to other suspects are eliminated from the officer’s thinking. Anyone,
police officer, counsel or judge can become infected by this virus.
Kennedy (2004), a defence lawyer who has assisted with the exoneration of several wrongfully
convicted individuals, argued that “police misconduct, which takes the form of overzealous and
shoddy investigative practices, and specifically “tunnel vision” (p. 200), is the most significant
factor leading to wrongful convictions. To this, Kennedy added:
Tunnel vision can result from incompetence, negligence, or simply a desire to
secure a conviction at any and all costs. It may occur when a police officer,
Crown counsel, or judge believes, prior to the presentation of all the evidence,
that the defendant is guilty… Tunnel vision leads to the use of worthless
evidence, the consideration of erroneous forensic science, and a reliance on the
testimony of jailhouse informants (pp. 200-201).
According to the FPT Committee (2004), tunnel vision is:
the single minded and overly narrow focus on an investigation or prosecutorial
theory so as to unreasonably colour the evaluation of information received and
one’s conduct in response to the information. Tunnel vision, and its perverse by-
Bounded Rationality and Criminal Investigations
23
product “noble cause corruption,” are the antithesis of the proper roles of the
police and Crown Attorney. Yet tunnel vision has been identified as a leading
cause of wrongful convictions in Canada and elsewhere (p. 35).
McFarlane (2004) suggested that tunnel vision occurs when the:
investigative team focuses prematurely, resulting in the arrest and prosecution of a
suspect against whom there is some evidence, while other leads and potential
lines of investigation go unexplored (p. 40).
Findley and Scott (2006) argued that tunnel vision is:
a natural human tendency that has particularly pernicious effects in the criminal
justice system. By tunnel vision, we mean that “compendium of common
heuristics and logical fallacies,” to which we are all susceptible, that lead actors in
the criminal justice system to “focus on a suspect, select and filter the evidence
that will ‘build a case’ for conviction, while ignoring or suppressing evidence that
points away from guilt.” The process leads investigators, prosecutors, judges, and
defence lawyers alike to focus on a particular conclusion and then filter all
evidence in the case through the lens provided by that conclusion. Through that
filter, all information supporting the adopted conclusion is elevated in
significance, viewed as consistent with the other evidence, and deemed relevant
and probative. Evidence inconsistent with the chosen theory is easily overlooked
or dismissed as irrelevant, incredible, or unreliable. Properly understood, tunnel
vision is more often the product of the human condition as well as institutional
and cultural pressures, than of maliciousness or indifference (p. 292).
Martin (2004) stated that tunnel vision is:
Bounded Rationality and Criminal Investigations
24
a set of preconceptions and heuristics that causes police investigators to select
evidence to build a case for the conviction of their chosen suspect while
suppressing or ignoring information and interpretations that point away from
guilt.
Lastly, Wilson (2003) stated that:
We have to remember that wrongful convictions do not happen out of malice. I
have never met a policeman who has deliberately tried to frame an innocent man
– this is not how wrongful convictions occur. Instead, investigators become
convinced of the guilt of people like Mr. Sophonow because the evidence itself
appears to be so convincing. It is crucial to remember that the case against an
accused, on the evidence, can be extremely compelling and yet the accused may
be innocent (p. 5).
Some of these definitions clearly imply police misconduct, as evidenced from the phrase
“noble cause corruption” in the FPT Committee’s report, where a conviction is knowingly
obtained under falsehoods or improper procedures because the police and/or prosecutor believe
the accused to be guilty. By observing in hindsight a case that ended in a wrongful conviction
and noticing that information was ignored during the investigation, it is not surprising that police
misconduct and tunnel vision might be viewed synonymously. Ignoring information certainly
could be the result of either misconduct or tunnel vision. But, it does not necessarily follow that
misconduct is to blame each and every time that an investigator ignores a piece of information.
It is our contention that these are two totally separate concepts that have been confounded.
Whereas malicious investigators may intentionally build a case against a suspect irrespective of
their guilt, detectives using tunnel vision really believe their primary suspect is responsible for
Bounded Rationality and Criminal Investigations
25
the crime. It is this belief that causes them to ignore some information – that which simply does
not register in their thoughts because it does not fit with the story they developed through their
interpretation of the evidence. Or if it does register, it might be “explained away” on the basis of
the investigator’s experience or the other evidence in the case. Tunnel vision, in our view,
guides information search in complex, uncertain, and pressure-filled situations. Ignoring some
information is an integral part of all heuristic strategies. Once it is agreed that misconduct and
tunnel vision are separate processes, we can begin to understand whether tunnel vision is really a
flawed cognitive strategy.
Despite the ambiguity associated with tunnel vision, there have been several
recommendations to remove tunnel vision from the investigative process (Cory, 2001; Findley &
Scott, 2006; FPT Committee, 2004; McFarlane, 2004). Such recommendations and concerns,
however, have been raised despite any compelling empirical evidence that tunnel vision is
maladaptive or even that it can be avoided. This type of recommendation is referred to as
“common-sense” based (Gendreau et al., 2002). Before such policy recommendations can be
implemented with any degree of confidence, they must be validated by systematic research.
Indeed, there are many questions about tunnel vision that remain unanswered. For example, how
often is tunnel vision used? When does tunnel vision lead to good decisions? When does tunnel
vision lead to bad decisions? Why does it work? Why does it fail? How often does tunnel
vision lead to successful criminal investigative outcomes? How often does tunnel vision lead to
criminal investigative failures? Is it realistic to expect the police to follow all lines of inquiry
even if they have a viable suspect identified? How many suspects should the police consider?
When should they stop searching for suspects? How much evidence is required to define a prime
Bounded Rationality and Criminal Investigations
26
suspect? How should police officers choose between equally viable suspects? Perhaps to start
we should ask: What exactly is tunnel vision?
Operationalizing Tunnel Vision
Even though tunnel vision has often been blamed for investigations going awry, the
concept has not been defined in a way that allows it to be meaningfully studied and scrutinized.
Cognitive strategies are normally defined in a way that clearly outlines the step-by-step process
(e.g., an algorithm) that leads to a decision. This has never been done for tunnel vision,
however.4 In this section, we will discuss a number of mental strategies that could potentially
comprise tunnel vision. Each heuristic may serve as a useful function in human cognition – and
in criminal investigations.
Judgment and decision making research provides at least four heuristics that may account
for a portion of the process that is commonly referred to as tunnel vision: (1) the satisficing
heuristic; (2) the take the best heuristic; (3) the elimination-by-aspects heuristic; and (4)
confirmation bias. Satisficing is a heuristic that may potentially be used to guide the search for a
primary suspect by ignoring some of the available information. The next two heuristics, which
also ignore some of the available information, can potentially be used to select a suspect from a
list of known suspects. Each of these three heuristics form part of the adaptive toolbox in the
bounded rationality framework and each involves rules for how to search for information, stop
that search, and make a decision. They have been proposed by bounded rationality researchers
as alternatives to rational2 models. In experiments where decisions made with these models are
compared to decisions made using complex models, the simple models perform at least as well
4 While Findley and Scott (2006) argue that tunnel vision is multidimensional and allude to the
idea that it is made up of heuristics, they draw upon vague cognitive tendencies in their attempt to operationalize tunnel vision.
Bounded Rationality and Criminal Investigations
27
and even better in some situations (e.g., Dhami, 2003, Gigerenzer, 2000). Confirmation bias is
involved during the building of a case against a suspect whom is believed to be the offender.
This is a psychological tendency that occurs when people attempt to support and defend their
decisions, but it can also be viewed as a component of the decision-making process. Arguably,
each of these psychological mechanisms (and other seemingly irrational strategies) serve
adaptive functions in human cognition. Evolutionary psychology maintains that, otherwise,
these mental tools would have been eliminated from our minds.
Satisficing, briefly discussed earlier, is a term that was introduced to the decision making
literature by Simon in the 1950s. It is actually a term that conveniently combines “satisfy” and
“suffice,” and essentially means looking for a good-enough option rather than the best option
when making a decision. The satisficing heuristic involves the sequential consideration of
options until one is found that appears to meet a certain aspiration level. Once that option is
chosen, no further options are considered. Within the context of a criminal investigation,
suspects are usually discovered one at a time and considered in that order. Police officers will
compare all relevant evidence on each suspect to the pre-defined level of evidence that, in the
police officers’ experience, is an indication of guilt. Any suspect that does not meet the
aspiration level is rejected and one primary suspect (the first suspect that meets all the aspiration
levels) emerges. If no suspects meet all the desired levels of evidence, the level of evidence
required may be relaxed and the process repeated, or the suspect who comes closest to the
aspiration level may be chosen. For example, a detective investigating a murder may determine
that the likely offender is a Caucasian male between 18 and 25 years of age, who lives near the
crime scene, has a previous conviction for burglary, and knows the victim. According to the
satisficing heuristic, the first suspect encountered that satisfies all these bits of information (not
Bounded Rationality and Criminal Investigations
28
simply the very first suspect who is encountered) will suffice as the primary suspect and the
search for other suspects is then ceased. In actuality, however, more than one suspect might
meet the aspiration level. Other heuristics might be used to choose the best suspect from that set
of possible suspects.
In the example about purchasing a house that was discussed earlier in this chapter, it can
be seen that satisficing may not always lead to the best possible choice. After you purchase a
house that meets all of your criteria, you may continue to browse the market only to find a better
house with a nicer view than your new home. This is a prime example of the type of situation
that is the basis of the rationality debate. Satisficing is a rational1 strategy that allows us to make
decisions in complex situations where there are so many factors involved that it would be
impossible to integrate them all into a decision. Yet using it might cause one to make a sub-
optimal (irrational2) choice. So a rational strategy produces an apparently irrational result. But,
satisficing allowed the home buyer to reach a good enough decision, which meant that the
process of buying a house could move forward. Therefore, using the satisficing heuristic aided
the achievement of the home buyer’s primary goal. Only in hindsight can it be learned that this
house may not have been the best choice. In criminal investigations, however, the situation is
different. There is a “right answer,” that is, there is normally just one suspect who is responsible
for a crime in question. But police officers probably use this heuristic nonetheless. It involves
setting standards and making sure those standards are met before coming to a final conclusion.
The standards are based on training and experience with similar crimes. This allows
investigators to make a decision and move the case forward. What better strategy is there? If the
evidence points to a suspect, and the police wrongly conclude the suspect is the rightful offender,
this may be an error but it is not an irrational conclusion because the detective followed the
Bounded Rationality and Criminal Investigations
29
evidence. In such a situation, it would be interesting to determine whether other investigators (or
anyone else) would also have followed the same line of inquiry and chosen the wrong suspect.
The take the best heuristic (TTB) makes decisions on the basis of just a few pieces of
data and ignores the rest of the available information (Gigerenzer, 2000; Gigerenzer, Hoffrage, &
Kleinbölting, 1991). The recognition heuristic, described earlier in this chapter, is a simplified
form of TTB in which decisions are based on recognition. Options that are not recognized are
eliminated as potential choices. Recall that the recognition heuristic is only used to make a final
decision when only one option is recognized. In other situations, recognition is used as a
preliminary cue to eliminate some options before additional information is considered.
Informational cues pertaining to the remaining options are then considered in a manner that
involves considering the “best” (most subjectively valid) cue first, then the second best, and so
on, until a cue is found that discriminates between the options.
For police officers assessing whether a case can be built against a suspect, these cues
might include motive, alibi, and prior criminal record, although the specific factors will vary
because there is an element of uniqueness in each case. It is assumed that the list of all potential
suspects is available prior to determining the primary suspect. The police would then evaluate
the motives of the potential suspects, eliminating any who had no motive. They would then do
the same for alibi, then prior criminal record, and other information, until only one suspect
remains. This description of TTB assumes that all cues are binary (present or not present), but
the heuristic can also be generalized in such a way that it considers the values of the cues
(Gigerenzer, 2000) such as the strength of the alibi or the degree of prior criminal activity. In
any case, TTB involves making a decision by only considering enough information to distinguish
Bounded Rationality and Criminal Investigations
30
between options. Some information is simply not used although it may be available, hence TTB
is often referred to by the slightly longer name, “take the best, ignore the rest.”
The elimination-by-aspects (EBA) heuristic, originally proposed by Tversky (1972), is
similar to TTB except that it considers the values of the cues (e.g., degree of prior criminal
activity). The most important piece of evidence is determined, a cut-off value is set, and
suspects are rejected if they do not surpass the cut-off level. This process is repeated with the
second most important piece of evidence, and so on, until only one suspect is left. For example,
the major crime investigator may have four possible suspects and they may all have alibis (which
the investigator believes to be the most important piece of evidence in the case), but one may
have a weaker alibi than the rest. Like TTB, this heuristic does not require considering all pieces
of information when choosing a primary suspect.
Once a primary suspect has been selected, the objective of an investigator is to build a
case against that person for the prosecution. Attempting to find information to support a belief is
commonly referred to as confirmation bias. Confirmation bias, by definition, is a type of
information search strategy whereby people look for information to confirm their beliefs,
theories, or hypotheses, and avoid or misinterpret contradictory evidence so that it does not
disconfirm their beliefs (Arkes & Harkness, 1980; Evans, 1989; Lewicka, 1998; Nickerson,
1998; Wason, 1960). The confirmation bias as it is currently explained in the literature, then,
refers to two separate mechanisms: (1) selective (confirmatory) information search; and (2)
biased interpretation of information (so that it does not disconfirm our beliefs).
Confirmation bias has been called both a “ubiquitous phenomenon” (Nickerson, 1998)
and the primary bias (Klein, 2001). These authors have observed that all people have a tendency
to search for information that confirms their beliefs – a tendency that appears to play a role in
Bounded Rationality and Criminal Investigations
31
many aspects of human cognition. Indeed, the scientific method is designed as an attempt to
overcome confirmation bias – by teaching scientists to seek evidence that disconfirms theories.
Klein (2001), however, noted that one of the most common strategies of scientific research is to
derive a prediction from a theory and show that it is accurate, thereby strengthening the
reputation of that theory. Interviews with prominent and successful NASA scientists (Mitroff,
1974) revealed that a number of them were committed to confirming their theories and saw this
type of strategy as both desirable and necessary. These scientists claim that seeking
confirmatory evidence is necessary for the development and refinement of new theories. Other
researchers have also shown that scientists and other expert professionals do not always search
for disconfirmatory evidence (Greenwald, Pratkanis, Leippe, & Baumgardner, 1986; Haverkamp,
1993; Mahoney & DeMonbreun, 1977). Mahoney and DeMonbreun (1977), for instance,
revealed that the reasoning skills of 15 psychologists and 15 physical scientists (all with PhDs)
did not differ from the reasoning skills of 15 conservative Protestant ministers. When successful
scientists use confirmation seeking strategies, the strategies are perceived as a sign of persistence
or focused determination. Ironically, however, some scientists criticize their research subjects
for using similar strategies on judgment and decision-making tasks.
These arguments can similarly be applied to the criminal investigative context. Officers
who build a successful case by searching for evidence that supports their belief while ignoring
evidence that contradicts their belief would likely be applauded for being persistent, focused,
determined, and dedicated. Those who use the same strategy in a case that ends in a wrongful
conviction, on the other hand, might be accused of using confirmation bias (or perhaps tunnel
vision), as though they neglected to properly fulfill their duties. The confirmation bias is said to
violate the “fully rational” expectation that police officers should search for endless amounts of
Bounded Rationality and Criminal Investigations
32
evidence or falsifying evidence in a totally objective manner, weigh and combine it all, and then
make the best decision. It appears that the outcome of the investigation (rather than how well it
was conducted) is the primary determinant of whether or not the investigative process is
criticized.
If confirmation bias is a conscious strategy, then it should be possible to teach people not
to use it. However, people may seek confirmatory evidence unconsciously rather than
consciously. Mynatt, Doherty, and Tweney (1977, 1978) trained participants on a falsification
technique and an alternate hypotheses technique and found that it did not decrease the prevalence
of confirmatory seeking behaviour when testing hypotheses. The researchers also discovered a
counterintuitive finding – a few participants were actually led astray by the search for
disconfirming evidence (i.e., the falsification technique). Such research suggests that even
individuals who are explicitly trained to make decisions and test hypotheses in a fully rational
manner, using normative strategies, often fail to do so, and training people to use “optimal”
search strategies can sometimes result in worse decisions. Mynatt et al. further found that
promising, but partially incorrect, hypotheses were quickly abandoned when disconfirmation was
received, and participants turned to other hypotheses that were much further from the solution
they were trying to find. So, while searching for confirmatory evidence might increase the
likelihood of supporting a prior belief, it does not completely rule out the possibility that
disconfirmatory evidence might also be discovered.
Tunnel vision appears to consist of a set of heuristics because, by definition, some of the
available information is ignored. There are many documented heuristics that people can use to
search for information, end that search, and make a decision. We have proposed just four that
may comprise the concept of tunnel vision, as currently used in the criminal justice literature, to
Bounded Rationality and Criminal Investigations
33
illustrate the importance of operationalizing concepts in a way that allows them to be empirically
tested. Experimental methods already exist for testing the four strategies we have presented,
whereas tunnel vision appears to be an ambiguous catchphrase that currently cannot be subjected
to the appropriate testing that would be required to determine its actual role in criminal
investigative successes and failures.
Conclusion
In our opinion, there is a gap between the reality of how police officers make decisions
and how the criminal justice system (and the general public) expects them to make decisions.
Given the uncertain, dynamic, and pressure-filled nature of criminal investigations and the
demands of the adversarial justice system, it is not reasonable to recommend that police officers
use “optimal” decision-making strategies. Just like a substantial number of psychological
researchers over the past several decades, those who cite tunnel vision as a cause of wrongful
convictions have made a very important oversight – investigative decisions are made by humans
in the real world, not by supercomputers in some ideal place where time, knowledge, and
resources are unlimited. Decision-making strategies that ignore information, including the
heuristics that comprise tunnel vision, are thus more psychologically and ecologically plausible
than those that strive for optimality.
Unfortunately, policy recommendations to eliminate wrongful convictions by eradicating
mental viruses are not based on any hard facts. Empirical research on the nature of tunnel vision
should be conducted before concluding it causes criminal investigative failures. Comparative
research is required to examine, for example, the prevalence of tunnel vision in criminal
investigations that resulted in both wrongful and rightful convictions. Only then can it be stated
Bounded Rationality and Criminal Investigations
34
with any certainty whether using tunnel vision is a poor way to approach investigative decision
making. Perhaps tunnel vision is used by investigators in every case, but only a very small
percentage of these result in wrongful convictions. This cannot be determined from the present
criminal justice literature because only investigative fiascos have been examined. That is not to
say that the analysis of investigative fiascos is not a useful starting point for identifying the
possible causes of wrongful convictions. Before the apparent use of tunnel vision is used to
drive policy recommendations, however, systematic research demonstrating the role of tunnel
vision in criminal investigative failures is needed.
Ironically, it seems to us that everyone uses tunnel vision, including those who have
argued that tunnel vision is a flawed mental strategy (by presenting only anecdotes that seem to
support their arguments). Some of the authors that we cite seem to have ignored research that
might disconfirm or weaken their arguments. Findley and Scott (2006), for example, made an
argument about tunnel vision that was similar to ours. Like us, they argued that tunnel vision
does not result from malice, and is comprised of heuristic processes. They made their argument
by focusing on the heuristics and biases literature, however, and appear to have ignored, or
perhaps did not search, the bounded rationality literature. The former generally claims that
heuristics are bad, while the latter argues their use can lead to smart decisions. Findley and Scott
then criticize police officers for using tunnel vision and recommend that officers think more
critically when making investigative decisions. This is where our views deviate from that of
many other researchers. By incorporating bounded rationality research into the picture, we have
argued that heuristics, including tunnel vision, might be adaptive processes that allow people to
function in a complex world. According to this view, asking police officers to avoid tunnel
vision might not be a feasible recommendation. Moreover, because tunnel vision helps police
Bounded Rationality and Criminal Investigations
35
officers achieve the positive goal of remaining focused, avoiding tunnel vision may not even be
desirable.
We admit that even we used tunnel vision when we wrote this chapter. Our approach
was to come to a conclusion regarding the use of tunnel vision through our psychological and
legal knowledge and then proceed to construct a rationale for these conclusions by searching for
evidence that supported them. But we were not intentionally trying to mislead anyone when we
chose not to include arguments from some sources. It is simply not feasible to incorporate every
argument that could potentially have an impact on this chapter. We strongly believe,
furthermore, that the bounded rationality approach currently holds the most promise for
understanding human decision making. Of course, if you read this chapter and give it a positive
evaluation, then you will probably conclude that our approach was good, whereas if you do not
agree with our position, you will probably accuse us of being biased. Although we may be
criticized for doing so, we are openly admitting that we used a tunnel vision approach to build
our argument because: (1) our knowledge of this area allows us to believe that we are presenting
a valid argument; and (2) confirmatory strategies have been argued to serve an adaptive purpose
in human reasoning.
In conclusion, the main goal of this chapter is to stop the spread of the idea that heuristics
are inevitably flawed. We recommend that people always consider the positive objectives that
heuristics might serve before concluding that people are irrational, lazy, or mentally inept. We
hope this chapter will help contribute to the larger, ongoing counter-revolution aimed at
balancing the view of the role and value of heuristics in human decision making.
Bounded Rationality and Criminal Investigations
36
References
Anderson, B., & Anderson, D. (1998). Manufacturing guilt: Wrongful convictions in Canada.
Halifax, NS: Fernwood Publishing.
Arkes, H. R., & Harkness, A. R. (1980). The effect of making a diagnosis on subsequent
recognition of symptoms. Journal of Experimental Psychology: Human Learning and
Memory, 6, 568-575.
Atkinson, R. C., & Shiffrin, R. M. (1968). Human memory: A proposed system and its control
processes. In K. W. Spence & J. T. Spence (Eds.), The psychology of learning and
motivation (Vol. 2, 89-195). New York: Academic Press.
Baddeley, A. D. (1992). Working memory. Science, 255, 556-569.
Campbell, K., & Denov, M. (2004). The burden of innocence: Coping with a wrongful
imprisonment. Canadian Journal of Criminology and Criminal Justice, 46, 139-163.
Cory, P. (2001). Commission of Inquiry Regarding Thomas Sophonow. Manitoba Justice,
Province of Manitoba. http://www.gov.mb.ca/justice/sophonow/index.html.
Crego, J., & Alison, L. (2004). Control and legacy as functions of perceived criticality in major
incidents. Journal of Investigative Psychology and Offender Profiling, 1, 207-225.
Czerlinski, J., Gigerenzer, G., & Goldstein, D. G. (1999). How good are simple heuristics? In G.
Gigerenzer, P. M. Todd, & The ABC Research Group (Eds.), Simple heuristics that make us
smart (pp. 97-118). New York: Oxford University Press.
de Poot, C. J., & van Koppen, P. J. (2004). Police detectives at work: On the nature and
consequences of problem solving strategies in police investigations. Manuscript submitted
for publication.
Bounded Rationality and Criminal Investigations
37
Dhami, M. K. (2003). Psychological models of professional decision making. Psychological
Science, 14, 175–180.
Dhami, M. K., & Harries, C. (2001). Fast and frugal versus regression models of human
judgment. Thinking and Reasoning, 7, 5–27.
Eck, J. E. (1979). Managing case assignments: The burglary investigation decision model
replication. Washington, DC: Police Executive Research Forum.
Edwards, W. (1968). Conservatism in human information processing. In B. Kleinmuntz (Ed.),
Formal representation of human judgment (pp. 17-52). New York: Wiley.
Edwards, W., Lindman, H., & Savage, L. J. (1963). Bayesian statistical inference for
psychological research. Psychological Review, 70, 193-242.
Epp, J. A. (1997). Penetrating police investigative practice post-Morin. UBC Law Review, 31,
95-126.
Evans, J. St.B. T. (1989). Bias in human reasoning: Causes and consequences. Hove, UK:
Laurence Erlbaum.
FPT Heads of Prosecutions Committee Working Group. (2004). Report on the prevention of
miscarriages of justice. Ottawa: Department of Justice.
Findley, K. A., & Scott, M. S. (2006). The multiple dimensions of tunnel vision in criminal
cases. Wisconsin Law Review, 2, 291-397.
Fiske, S. T., & Taylor, S. E. (1991). Social cognition. New York: McGraw-Hill.
Forst, B. (2004). Errors of justice: Nature, sources and remedies. New York, NY: Cambridge
University Press.
Gendreau, P., Goggin, C., Cullen, F. T., & Paparozzi, M. (2002). The common-sense revolution
and correctional policy. In J. Maguire (Ed.), Offender rehabilitation and treatment: Effective
Bounded Rationality and Criminal Investigations
38
programmes and policies to reduce re-offending (pp. 359–386). Chichester, UK: John Wiley
and Sons.
Gigerenzer, G. (2000). Adaptive thinking: Rationality in the real world. New York: Oxford
University Press.
Gigerenzer, G. (2001). The adaptive toolbox. In G. Gigerenzer & R. Selton (Eds.), Bounded
rationality: The adaptive toolbox (pp. 37-50). Cambridge, MA: The MIT Press.
Gigerenzer, G., Hoffrage, U., & Kleinbolting, H. (1991). Probabilistic mental models: A
Brunswikian theory of confidence. Psychological Review, 98, 506-528.
Gigerenzer, G., & Selton, R. (2001). Bounded rationality. London: MIT Press.
Gigerenzer, G., & Todd, P. M. (1999). Fast and frugal heuristics: The adaptive toolbox. In G.
Gigerenzer, P. M. Todd, & the ABC Research Group (Eds.), Simple heuristics that make us
smart (pp. 3-34). New York: Oxford University Press.
Gigerenzer, G., Todd, P., & The ABC Research Group. (1999). Simple heuristics that make us
smart. New York: Oxford University Press.
Gilovich, T., & Griffin, D. (2002). Heuristics and biases: Then and now. In T. Gilovich, D.
Griffin, & D. Kahneman (Eds.), Heuristics and biases: The psychology of intuitive judgment
(pp. 1-18). Cambridge, UK: Cambridge University Press.
Gilovich, T., Griffin, D., & Kahneman, D. (2002). Heuristics and biases: The psychology of
intuitive judgment. Cambridge, UK: Cambridge University Press.
Goff, C. (2001). Criminal justice in Canada. Scarborough, ON: Nelson Thomson Learning.
Goldberg, L. R. (1965). Diagnosticians vs. diagnostic signs: The diagnosis of psychosis vs.
neurosis from the MMPI. Psychological Monographs, 79 (9, Whole No. 602).
Bounded Rationality and Criminal Investigations
39
Goldstein, D. G., & Gigerenzer, G. (1999). The recognition heuristic: How ignorance makes us
smart. In G. Gigerenzer, P. M. Todd, & the ABC Research Group (Eds.), Simple heuristics
that make us smart (pp. 37-58). London: Oxford University Press.
Goldstein, D. G., & Gigerenzer, G. (2002). Models of ecological rationality: The recognition
heuristic. Psychological Review, 109, 75-90.
Goldstein, W. M., & Hogarth, R. M. (1997). Judgment and decision research: Some historical
context. In W. M. Goldstein & R. M. Hogarth (Eds.), Research on judgment and decision
making: Currents, connections, and controversies (pp. 3-65). Cambridge: Cambridge
University Press.
Greenwald, A. G., Pratkanis, A. R., Leippe, M. R., & Baumgardner, M. H. (1986). Under what
conditions does theory obstruct research progress? Psychological Review, 93, 216-229.
Grounds, A. (2004). Psychological consequences of wrongful conviction and imprisonment.
Canadian Journal of Criminology and Criminal Justice, 46, 165-182.
Haverkamp, B. E. (1993). Confirmatory bias in hypothesis testing for client-identified and
counselor self-generated hypotheses. Journal of Counseling Psychology, 40, 303-315.
Huff, C. R. (2004). Wrongful convictions: The American experience. Canadian Journal of
Criminology and Criminal Justice, 46, 107-120.
Huff, C. R., Rattner, A., & Sagarin, E. (1986). Guilty until proven innocent: Wrongful conviction
and public policy. Crime and Delinquency, 32, 518-544.
Innes, M. (2002). The ‘process structures’ of police homicide investigations. British Journal of
Criminology, 42, 669-688.
Kahneman, D. (1973). Attention and effort. New Jersey: Prentice-Hall.
Bounded Rationality and Criminal Investigations
40
Kahneman, D., Slovic, P., & Tversky, A. (1982). Judgment under uncertainty: Heuristics and
biases. Cambridge, UK: Cambridge University Press.
Kahneman, D., & Tversky, A. (1972). Subjective probability: A judgment of representativeness.
Cognitive Psychology, 3, 430-454.
Kahneman, D., & Tversky, A. (1973). On the psychology of prediction. Psychological Review,
80, 237-251.
Kennedy, J. P. (2004). Writing the wrongs: The role of defense counsel in wrongful convictions
– A commentary. Canadian Journal of Criminology and Criminal Justice, 46, 197-208.
Keppel, R. D., & Weis, J. G. (1994). Time and distance as solvability factors in murder cases.
Journal of Forensic Sciences, 49, 386-401.
Klein, G. (2001). Sources of power: How people make decisions. Cambridge, MA: The MIT
Press.
Krueger, J. I., & Funder, D. C. (2004). Towards a balanced social psychology: Causes,
consequences and cures for the problem-seeking approach to social behavior and cognition.
Behavioral and Brain Sciences, 27, 313-376.
Lerner, C. S. (2005). Reasonable suspicion and mere hunches. George Mason Law and
Economics Research Paper No. 05-20.
Lewicka, M. (1998). Confirmation bias: Cognitive error or adaptive strategy of action control? In
M. Kofta, G. Weary, & G. Sedek (Eds.), Personal control in action: Cognitive and
motivational mechanisms (pp. 233-258). New York, NY: Plenum Press.
Lyons, A., & Truzzi, M. (1991). The blue sense: Psychic detectives and crimes. New York: The
Mysterious Press.
Bounded Rationality and Criminal Investigations
41
MacFarlane, B. A. (2004). Convicting the innocent: A triple failure of the justice system.
Retrieved from http://wrongfulconviction.ca/Papers.htm.
Maguire, M. (2003). Criminal investigation and crime control. In T. Newburn (Ed.), Handbook
of policing. Devon, UK: Willan.
Mahoney, M. J., & DeMonbreun, B. G. (1977). Psychology of the scientist: An analysis of
problem-solving bias. Cognitive Therapy and Research, 1, 229-238.
Manktelow, K. (1999). Reasoning and thinking. Hove, Sussex. The Psychology Press.
Manning, P. K. (1977). Police work: The social organization of policing. Cambridge, MA: The
MIT Press.
Martin, D. L. (2004). Lessons about justice from the “laboratory” of wrongful convictions:
Tunnel vision, the construction of guilt and informer evidence. The Canadian Review of
Policing Research. Retried December 13, 200, from
http://crpr.icaap.org/index.php/crpr/index.
Meehl, P. E. (1954). Clinical versus statistical prediction. Minneapolis: University of Minnesota
Press.
Miller, G. A. (1956). The magic number seven, plus or minus two: Some limits on our capacity
for processing information. Psychological Review, 63, 81-93.
Mitroff, I. (1974). The subjective side of science. Amsterdam: Elsevier.
Mouzos, J., & Muller, D. (2001). Solvability factors of homicide in Australia: An exploratory
analysis. Trends and Issues in Crime and Criminal Justice (no. 216). Australian Institute of
Criminology, Canberra.
Myers, D. G. (2002). Intuition: Its powers and perils. New Haven: Yale University Press.
Bounded Rationality and Criminal Investigations
42
Mynatt, C. R., Doherty, M. E., & Tweney, R. D. (1977). Confirmation bias in a simulated
research environment: An experimental study of scientific inference. Quarterly Journal of
Experimental Psychology, 29, 85-95.
Mynatt, C. R., Doherty, M. E., & Tweney, R. D. (1978). Consequences of confirmation and
disconfirmation in a simulated research environment. Quarterly Journal of Experimental
Psychology, 30, 395-406.
Nickerson, R. S. (1998). Confirmation bias: A ubiquitous phenomenon in many guises. Review
of General Psychology, 2, 175-220.
Nisbett, R. E., & Ross, L. (1980). Human inference: Strategies and shortcomings of social
judgment. Englewood-Cliffs: Prentice-Hall.
Orasanu, M., & Salas, D. (1993). Team decision making in complex environments, In G. A.
Klein, J. Orasanu, R. Calderwood, & C. E. Zsambok (Eds.), Decision making in action:
Models and methods (pp. 327-345). Norwood, NJ: Ablex Publishers.
Piattelli-Palmarini, M. (1994). Inevitable illusions. How mistakes of reason rule our minds. New
York: John Wiley & Sons.
Rosen, P. (1992). Wrongful convictions in the criminal justice system. Ottawa, ON: Research
Branch of the Library of Parliament.
Sanders, W. B. (1977). Detective work: A study of criminal investigations. New York: Free Press
Sawyer, J. (1966). Measurement and prediction, clinical and statistical. Psychological Bulletin,
66, 178-200.
Scullion, K. (2004). Wrongful convictions and the criminal conviction review process pursuant
to Section 696.1 of the Criminal Code of Canada. Canadian Journal of Criminology and
Criminal Justice, 46, 189-195.
Bounded Rationality and Criminal Investigations
43
Simon, H. A. (1955). A behavioral model of rational choice. Quarterly Journal of Economics,
69, 99-118.
Simon, H. A. (1956). Rational choice and the structure of the environment. Psychological
Review, 63, 129-138.
Smith, L., & Gilhooly, K. (2006). Regression versus fast and frugal models of decision-making:
The case of prescribing for depression. Applied Cognitive Psychology, 20, 265-274.
Snook, B., & Cullen, R. M. (2006). Recognizing national hockey league greatness with an
ignorance-based heuristic. Canadian Journal of Experimental Psychology, 60, 33-43.
Snook, B., Taylor, P. J., & Bennell, C. (2004). Geographic profiling: The fast, frugal, and
accurate way. Applied Cognitive Psychology, 18, 105-121.
Sporer, S. L. (1996). Psychological aspects of person descriptions. In S. L. Sporer, R. S.
Malpass, & G. Köehnken (Eds.), Psychological issues in eyewitness identification (pp. 53-
86). Mahwah, NJ: Earlbaum.
Todd, P. M., & Gigerenzer, G. (2000). Précis of Simple heuristics that make us smart.
Behavioral and Brain Sciences, 23, 727-780.
Todd, P. M., & Gigerenzer, G. (2003). Bounding rationality to the world. Journal of Economic
Psychology, 24, 143-165.
Tversky, A. (1972). Elimination by aspects: A theory of choice. Psychological Review, 79, 281-
299.
Tversky, A., & Kahneman, D. (1971). Belief in the law of small numbers. Psychological
Bulletin, 76, 105-110.
Tversky, A., & Kahneman, D. (1973). Availability: A heuristic for judging frequency and
probability. Cognitive Psychology, 5, 207-232.
Bounded Rationality and Criminal Investigations
44
Tversky, A., & Kahneman, D. (1974). Judgment under uncertainty: Heuristics and biases.
Science, 185, 1124-1131.
Tversky, A., & Kahneman, D. (2002). Extensional versus intuitive reasoning: The conjunction
fallacy in probability judgment. In T. Gilovich, D. Griffin, & D. Kahneman (Eds.),
Heuristics and biases: The psychology of intuitive judgment (pp. 19-48). Cambridge, UK:
Cambridge University Press.
Young, A. N. (1996). The Charter, the Supreme Court of Canada, and the constitutionalization of
the investigative process. In J. Cameron (Ed.), The charter’s impact on the criminal justice
system (pp. 1-33). Scarborough, ON: Thomson Canada Limited.
Wason, P. C. (1960). On the failure to eliminate hypotheses in a conceptual task. Quarterly
Journal of Experimental Psychology, 12, 129-140.
Weisman, R. (2004). Showing remorse: Reflections on the gap between expression and
attribution in cases of wrongful conviction. Canadian Journal of Criminology and Criminal
Justice, 46, 121-138.
Williams, L. J. (1985). Tunnel vision induced by a foveal load manipulation. Human Factors,
27, 221-227.
Wilson, P. J. (2003). Wrongful conviction: Lessons learned from the Sophonow Public Inquiry.
Canadian Police College.
Bounded Rationality and Criminal Investigations
45
Authors’ Note
Correspondence: Brent Snook, Psychology Department, Science Building, Memorial
University of Newfoundland, St. John’s, Newfoundland, Canada, A1B 3X9. Telephone: (709)
737-3101; facsimile: (709) 737-2430; e-mail: [email protected].
Support for the research reported in this paper was provided to the first author by the
Natural Sciences and Engineering Research Council of Canada, and to the second author by the
Social Sciences and Humanities Research Council of Canada.
Biographies
Brent Snook is an Assistant Professor and Director of the Bounded Rationality and the
Law Laboratory at Memorial University of Newfoundland. He obtained his PhD at the
University of Liverpool, United Kingdom. Brent’s research primarily involves the study of
bounded rationality and the law. He is particularly interested in the types of simple heuristics
that people use to make consequential decisions in legal settings, when and why those heuristics
work succeed and fail, and the conditions under which simple heuristics are used. His research
interests also include pseudoscience in the criminal justice system.
Richard M. Cullen is a graduate student at Memorial University of Newfoundland. He
obtained his BSc (Honours) from Memorial in 2005. His general research interests include
bounded rationality, forensic psychology, and social cognition. Richard’s recent research
involves the development of a model that explains how people form beliefs about investigative
techniques in the absence of compelling empirical evidence. His graduate work involves
studying evidence for the usage of simple heuristics in legal decision making.